Bubbles, Crashes, and Circuit Breakers: When Markets Go Wild (Chapter 28)
Previous: Floor vs Automated Trading
Chapter 28 is the most dramatic chapter in Trading and Exchanges. It covers the moments when markets go completely sideways. Bubbles that inflate until they pop. Crashes that destroy enormous wealth in hours. And the regulatory responses that try to prevent it all from happening again.
Harris approaches the subject with a microstructure lens, which makes his analysis different from most crash retrospectives. He is not just asking “what happened” but “how did market structure contribute to what happened, and can we design better structures?”
How Bubbles Form
The bubble pattern is surprisingly consistent across centuries. It starts with genuinely good news about fundamentals. A new technology. A promising market. Traders get excited and buy. The initial price increase is probably justified.
But then the second wave hits. Momentum traders pile in because they see prices going up and want the same gains their friends are getting. Order anticipators buy in front of expected uninformed buyers. The combined pressure pushes prices above fundamental values.
Here is the thing that makes bubbles so hard to pop early: value traders might recognize the overpricing but cannot or will not stop it. They might not have enough capital to carry large short positions. They might not be able to sell short easily. They might lack confidence in their own value estimates, especially with unproven technologies. And even if they are right about values, they face brutal timing risk. Sell short too early and you get crushed as uninformed traders push prices even higher. Wait too long and other value traders take the profits.
Harris nails this psychological dynamic. Trading against the majority requires enormous courage. Markets generally aggregate information well, so any contrarian has to honestly ask: why do I think I know better than everyone else? In a bubble, value traders need to be confident in their analysis while simultaneously accepting that the majority of traders disagree.
The Crash Dynamics
Eventually sellers get aggressive. Maybe long-term holders want to lock in gains. Maybe bad fundamental news arrives. Once prices start falling, the whole machine runs in reverse. Momentum buyers disappear. Overly optimistic buyers lose confidence. Late buyers panic and sell to stop their losses. Margin calls force leveraged traders to sell. Stop loss orders trigger. Order anticipators sell ahead of the anticipated liquidations.
And then prices can overshoot on the downside. Panicked sellers demand so much liquidity that prices drop below fundamental values. This creates the “dead cat bounce,” a temporary recovery when traders realize the market overreacted.
Harris makes an important distinction between fundamental and transitory volatility. Fundamental volatility is caused by actual news and has permanent effects. Transitory volatility is caused by uninformed trading pressure and tends to reverse. Most crashes involve both. The bubble creates transitory overpricing. The crash corrects it, but panic selling can create additional transitory underpricing.
The Crash Hall of Fame
Harris walks through several major crashes, and each one teaches different lessons about market structure.
The 1929 Crash. The Dow dropped 13% on October 28, then another 12% on October 29. The bubble was fueled by excessive speculation on margin in new technology stocks (radio companies, the “internet stocks” of their day). RCA did not make a new high for 34 years. The crash led to the Fed getting authority over stock margins, and they immediately set them at 45%. Harris notes that while people associate the crash with the Great Depression, most economists attribute the Depression to bad monetary policy and banking failures, not the crash itself.
The 1987 Crash. This is the centerpiece of the chapter, and Harris gives it the detailed treatment it deserves. The Dow lost 23% on Monday, October 19, 1987. It remains the largest single-day percentage loss ever.
The 1987 crash is a perfect case study in how market structure can amplify problems. The main culprit was portfolio insurance, a dynamic trading strategy that replicates a put option by selling stocks as they fall and buying as they rise. It is essentially the same as a massive collection of stop loss orders.
Portfolio insurance works fine when the money involved is small. But by October 1987, roughly 100 billion dollars was subject to portfolio insurance strategies. Harris shows that a 1% market drop would require portfolio insurers to sell about 10 million shares, against daily average volume of only 160 million shares. That is an enormous amount of one-directional selling pressure.
Making it worse: everyone knew about the portfolio insurance problem but nobody knew exactly how much money was under the strategy. Several academics had written papers about it. It was discussed at conferences. But the uncertainty about total exposure meant nobody wanted to step in and buy until they believed all the selling was done.
The crash was compounded by capacity failures at the NYSE. Dot matrix printers on the floor could not print SuperDot orders fast enough. Some printers broke. Orders sat in queues for over an hour. Traders had no idea if their orders had executed, so many placed duplicate sell orders by phone.
The index futures market, which was faster, led the cash market down. The spread between cash and futures widened to more than 10%. Arbitrageurs who normally keep these markets aligned stopped trading because they could not get reliable executions in the stock market. The disconnection between markets added to the confusion and panic.
And on Nasdaq, many dealers simply took their phones off the hook because they did not want to trade. Traders who could not reach their dealers while watching the market collapse must have been terrified.
The 1989 Mini-Crash. The market dropped 7% on a Friday afternoon after banks announced they would not finance the UAL leveraged buyout. What made this crash interesting was that many traders had left early because of good weather. The market was unusually illiquid when the news hit. Harris mentions an unconfirmed but intriguing theory that some traders may have deliberately bluffed the market down, recognizing it was vulnerable. The market recovered within weeks.
The Nasdaq Bubble. The Nasdaq bubble of the late 1990s echoed 1929. Traders were wildly optimistic about internet and technology companies. Momentum investors poured money into undiversified tech mutual funds. These funds invested the new money into the same stocks they held, pushing prices higher, generating more returns, attracting more money. Internet trading platforms brought timid retail investors into the market who previously were intimidated by talking to brokers on the phone.
Circuit Breakers: Do They Work?
Harris examines several types of circuit breakers that regulators use to manage extreme volatility.
Trading halts stop trading when prices move by a specified amount. They might help if the volatility is caused by uninformed panic because they give people time to calm down and think. They can attract new liquidity suppliers who learn about the opportunity during the halt. And they switch the pricing mechanism from discriminatory (each limit order trades at its limit price) to uniform (everyone gets the same price at the reopening auction), which encourages more limit orders.
But trading halts can also backfire. The “gravitational effect” means traders might rush to submit orders before a halt kicks in, actually increasing volatility. And if value traders know the media will alert them when a halt happens, they might pay less attention between halts, making the market less liquid normally.
Price limits work similarly to trading halts but cap prices rather than stopping trading entirely. Harris includes a clever example showing how price limits can actually extract more margin from traders who are heading toward bankruptcy. A trader who would know he is bankrupt after a 25-point drop might keep posting margin through five consecutive 5-point limit-down days, ultimately paying more to his broker than he would have in a single crash.
Transaction taxes and increased margins reduce trading, but they fall on everyone equally. They hurt dealers, informed traders, and utilitarian traders just as much as they hurt the gamblers and front-runners who actually cause problems. Harris argues that ideally you would restrict only the destabilizing traders, but in practice it is impossible to distinguish them from everyone else.
The Politics of Regulation
The last section of the chapter is brilliantly cynical. Harris lays out a game theory analysis of why regulators adopt the rules they adopt.
After the 1987 crash, the public demanded action. Harris shows that regulators faced a classic asymmetric incentive: if they adopt circuit breakers and another crash happens, people will not blame them for trying. If they do nothing and another crash happens, they will be held responsible. If they adopt mild circuit breakers and nothing happens, they have not done any harm. If they adopt severe circuit breakers and nothing happens, people will blame them for overreacting.
The result? Regulators rationally chose mild circuit breakers. The initial Rule 80B trading halt would only trigger at a 12% drop, something that had happened exactly once in US history. It was effectively a rule that said “we did something” without actually doing much.
Harris also explains NYSE Rule 80A, which restricted index arbitrage, as a case of regulatory capture. Specialists benefited from limiting arbitrageurs who competed with them. Since specialists had far more political power at the NYSE than arbitrageurs did, the rule was adopted. It probably slightly increased transitory volatility by preventing arbitrageurs from moving liquidity between markets.
The Real Lesson
The best way to prevent bubbles is not circuit breakers. It is empowering value traders who can recognize and trade against overpricing. That means making high-quality information cheap and accessible. It means removing barriers to short selling. It means letting the people who actually understand values act on their knowledge.
Most regulatory responses to crashes address symptoms rather than causes. They try to slow down the crash rather than prevent the bubble that created it. The conditions that lead to crashes usually build up over months or years. By the time the crash happens, the damage is already done.
This post is part of a series on Trading and Exchanges: Market Microstructure for Practitioners by Larry Harris (Oxford University Press, 2003, ISBN: 0-19-514470-8). This retelling covers Chapter 28.